Combination of remote sensing with crop modeling using Bayesian inferences to predict irrigated cotton yield
The primary goal of this study was to create a Bayesian framework that would incorporate remote sensing data to automatically calibrate the AquaCrop model for simulating cotton responses to irrigation strategies in the northern border of the United States Cotton Belt, which faces a lack of observati...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-08-01
|
| Series: | Agricultural Water Management |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S0378377425003890 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850075395925213184 |
|---|---|
| author | Farzam Moghbel Forough Fazel Jonathan Aguilar Nathan Howell Juan Enciso |
| author_facet | Farzam Moghbel Forough Fazel Jonathan Aguilar Nathan Howell Juan Enciso |
| author_sort | Farzam Moghbel |
| collection | DOAJ |
| description | The primary goal of this study was to create a Bayesian framework that would incorporate remote sensing data to automatically calibrate the AquaCrop model for simulating cotton responses to irrigation strategies in the northern border of the United States Cotton Belt, which faces a lack of observational data. Multiple regression models (linear and non-linear) were fitted to establish a correlation between cotton canopy cover (CC) values and aerial vegetation indices (EVI, EVI2, MACARI, NDRE, NDVI, NDSVI, OSAVI, and VARI) obtained from sUAS multispectral imagery for 2021 and 2022 growing seasons. The highest correlation was found between RGB-Based VARI index and cotton CC by fitting the linear model (R2 = 0.83 and RMSE = 0.12), which contradicted the results of other studies that emphasized the importance of using red-edge and near-infrared for monitoring crop canopy cover. A considerably less accurate correlation was detected for fitting the polynomial model (0.4 <RMSE<4.2). Furthermore, the MCARI index was found unsuitable for cotton monitoring under water stress conditions. Afterward, the Bayesian theorem-based Generalized Likelihood Uncertainty Estimation (GLUE) algorithm was linked to AquaCrop in the R environment to calibrate the model based on the remotely sensed CCs by seeking posterior distributions of the parameters through the Monte Carlo approach. Then, the model was validated for its key outputs, including cotton biomass, yield, and soil water content. The simulated CC results showed the model's automatic calibration success. The best performances of the model were found for simulating cotton biomass under 70 % and 80 % deficit irrigation conditions Pe = 0.88 % and −0.38 %) in 2022 and full irrigation conditions in 2021 Pe = 3.19 %); however, the biomass simulations were satisfactorily under all irrigation conditions. The outstanding performance of the AquaCrop was confirmed for reproducing cotton yield values regardless of irrigation conditions. The accurate retrieval of soil water dynamics by the model can introduce the framework created in this study as a robust tool to derive soil water content at cotton rootzone by having access to aerial RGB images. Overall, the findings of this study revealed that by supplying the introduced framework with one seasonal remote sensing data, the AquaCrop could successfully be turned into a decision support system for exploring irrigation scheduling strategies for producers of cotton in Kansas. |
| format | Article |
| id | doaj-art-c5b4ed138a674ea98dc9794a08b65eeb |
| institution | DOAJ |
| issn | 1873-2283 |
| language | English |
| publishDate | 2025-08-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Agricultural Water Management |
| spelling | doaj-art-c5b4ed138a674ea98dc9794a08b65eeb2025-08-20T02:46:19ZengElsevierAgricultural Water Management1873-22832025-08-0131710967510.1016/j.agwat.2025.109675Combination of remote sensing with crop modeling using Bayesian inferences to predict irrigated cotton yieldFarzam Moghbel0Forough Fazel1Jonathan Aguilar2Nathan Howell3Juan Enciso4Southwest Research–Extension Center, Kansas State University, 4500 E. Mary St., Garden City, KS 67846, USA; Biological and Agricultural Engineering Department, Kansas State University, Seaton Hall, Martin Luther King Jr. Drive, Manhattan, KS 66506, USA; Correspondence authors at: Southwest Research–Extension Center, Kansas State University, 4500 E. Mary St., Garden City, KS 67846, USA.Southwest Research–Extension Center, Kansas State University, 4500 E. Mary St., Garden City, KS 67846, USA; Biological and Agricultural Engineering Department, Kansas State University, Seaton Hall, Martin Luther King Jr. Drive, Manhattan, KS 66506, USA; Correspondence authors at: Southwest Research–Extension Center, Kansas State University, 4500 E. Mary St., Garden City, KS 67846, USA.Southwest Research–Extension Center, Kansas State University, 4500 E. Mary St., Garden City, KS 67846, USA; Biological and Agricultural Engineering Department, Kansas State University, Seaton Hall, Martin Luther King Jr. Drive, Manhattan, KS 66506, USACollege of Engineering, West Texas A&M University, Canyon, TX, 79016, USATexas A&M AgriLife Research, Texas A&M University, Weslaco, TX 78596, USA; Department of Biological and Agricultural Engineering, Texas A&M University, College Station, TX 77843, USAThe primary goal of this study was to create a Bayesian framework that would incorporate remote sensing data to automatically calibrate the AquaCrop model for simulating cotton responses to irrigation strategies in the northern border of the United States Cotton Belt, which faces a lack of observational data. Multiple regression models (linear and non-linear) were fitted to establish a correlation between cotton canopy cover (CC) values and aerial vegetation indices (EVI, EVI2, MACARI, NDRE, NDVI, NDSVI, OSAVI, and VARI) obtained from sUAS multispectral imagery for 2021 and 2022 growing seasons. The highest correlation was found between RGB-Based VARI index and cotton CC by fitting the linear model (R2 = 0.83 and RMSE = 0.12), which contradicted the results of other studies that emphasized the importance of using red-edge and near-infrared for monitoring crop canopy cover. A considerably less accurate correlation was detected for fitting the polynomial model (0.4 <RMSE<4.2). Furthermore, the MCARI index was found unsuitable for cotton monitoring under water stress conditions. Afterward, the Bayesian theorem-based Generalized Likelihood Uncertainty Estimation (GLUE) algorithm was linked to AquaCrop in the R environment to calibrate the model based on the remotely sensed CCs by seeking posterior distributions of the parameters through the Monte Carlo approach. Then, the model was validated for its key outputs, including cotton biomass, yield, and soil water content. The simulated CC results showed the model's automatic calibration success. The best performances of the model were found for simulating cotton biomass under 70 % and 80 % deficit irrigation conditions Pe = 0.88 % and −0.38 %) in 2022 and full irrigation conditions in 2021 Pe = 3.19 %); however, the biomass simulations were satisfactorily under all irrigation conditions. The outstanding performance of the AquaCrop was confirmed for reproducing cotton yield values regardless of irrigation conditions. The accurate retrieval of soil water dynamics by the model can introduce the framework created in this study as a robust tool to derive soil water content at cotton rootzone by having access to aerial RGB images. Overall, the findings of this study revealed that by supplying the introduced framework with one seasonal remote sensing data, the AquaCrop could successfully be turned into a decision support system for exploring irrigation scheduling strategies for producers of cotton in Kansas.http://www.sciencedirect.com/science/article/pii/S0378377425003890AquaCropBayesianCalibrationCottonDeficit IrrigationRemote Sensing |
| spellingShingle | Farzam Moghbel Forough Fazel Jonathan Aguilar Nathan Howell Juan Enciso Combination of remote sensing with crop modeling using Bayesian inferences to predict irrigated cotton yield Agricultural Water Management AquaCrop Bayesian Calibration Cotton Deficit Irrigation Remote Sensing |
| title | Combination of remote sensing with crop modeling using Bayesian inferences to predict irrigated cotton yield |
| title_full | Combination of remote sensing with crop modeling using Bayesian inferences to predict irrigated cotton yield |
| title_fullStr | Combination of remote sensing with crop modeling using Bayesian inferences to predict irrigated cotton yield |
| title_full_unstemmed | Combination of remote sensing with crop modeling using Bayesian inferences to predict irrigated cotton yield |
| title_short | Combination of remote sensing with crop modeling using Bayesian inferences to predict irrigated cotton yield |
| title_sort | combination of remote sensing with crop modeling using bayesian inferences to predict irrigated cotton yield |
| topic | AquaCrop Bayesian Calibration Cotton Deficit Irrigation Remote Sensing |
| url | http://www.sciencedirect.com/science/article/pii/S0378377425003890 |
| work_keys_str_mv | AT farzammoghbel combinationofremotesensingwithcropmodelingusingbayesianinferencestopredictirrigatedcottonyield AT foroughfazel combinationofremotesensingwithcropmodelingusingbayesianinferencestopredictirrigatedcottonyield AT jonathanaguilar combinationofremotesensingwithcropmodelingusingbayesianinferencestopredictirrigatedcottonyield AT nathanhowell combinationofremotesensingwithcropmodelingusingbayesianinferencestopredictirrigatedcottonyield AT juanenciso combinationofremotesensingwithcropmodelingusingbayesianinferencestopredictirrigatedcottonyield |